ML Data Scientist

Job Summary

We're looking for an ML Data Scientist to join our Geospatial Data team. You'll take ownership of building and improving machine learning models that analyze satellite imagery to support agricultural and environmental applications. The role involves running experiments, interpretingresults, and working through modeling challenges independently, while collaborating with scientists, product managers, and engineers to turn real-world use cases into reliable, production-ready solutions.

What You'll Do 

• Train, evaluate, and iteratively improve geospatial classification and regression models built on existing deep learning architectures using multi-spectral satellite time-series data.
• Analyze model performance across diverse geographies, seasons, and target classes, diagnosing failure modes and adjusting training configurations, sampling strategies, and hyperparameters to drive meaningful improvements.
• Design and execute experiments to validate modeling decisions, interpret results, and translate findings into actionable next steps.
• Work with domain scientists and subject matter experts to understand label quality, ground truth limitations, and how real-world complexity should inform modeling decisions.
• Collaborate cross-functionally with product managers, domain scientists, and engineering teams to translate use cases into robust ML solutions.
• Operate independently within an existing ML infrastructure stack, running training jobs on Vertex AI, managing data in BigQuery and Cloud Storage, and deploying models to production endpoints.
• Work with an engineering mindset to improve existing ML frameworks and infrastructure as needed to support new product needs.
• Implement active learning and automated QA workflows to continuously improve quality and model performance.
• Manage the full model lifecycle, including experiment tracking, model registration, versioning, and deployment to AI Platform endpoints.
• Contribute to data pipeline improvements and orchestration workflows where needed to unblock modeling work.

What We're Looking For

• 4-6 years of experience in ML engineering or data science, with at least 2 years working with remote sensing or geospatial data.
• Demonstrated ability to work independently, owning a modeling problem end-to-end, unblocking yourself, and delivering results with minimal hand-holding in a fast-moving environment.
• Strong intuition for model diagnostics, including interpreting training curves, confusion matrices, data leakage, and evaluation metrics to identify what's working and what isn't.
• Hands-on experience training deep learning models (TensorFlow or PyTorch) on multi-spectral time-series or imagery tasks, including distributed training (DDP, multi-GPU).
• Geospatial and remote sensing fundamentals, including projections/CRS, raster vs. vector, band math, cloud masking, time-series composites, and spatial statistics.
• Familiarity with SAR, multispectral, or hyperspectral sensor data.
• Comfortable operating in cloud ML infrastructure.
• MLOps fundamentals, including experiment tracking, dataset/model versioning, and basic CI/CD concepts.

Nice to Have

• Familiarity with workflow orchestration tools such as Argo Workflows, Airflow, Kubeflow, or similar.
• Familiarity with Google Earth Engine for large-scale geospatial data access and computation.
• Experience using AI coding assistants or agentic development tools as part of a day-to-day ML workflow, with an interest in applying agentic approaches to automate and accelerate the modeling lifecycle.